F1's brand-new regulatory era could not have kicked off with more chaos if it had been scripted. The 2026 Australian Grand Prix at Albert Park was billed as a fresh start for everyone — new power units, new aero rules, a genuinely reshuffled grid — and the pre-race prediction algorithm leaned hard into that uncertainty, crowning rookie Isack Hadjar as the man to beat and all but writing off the Mercedes and Ferrari drivers. Instead, Melbourne delivered one of the most predictions-shredding afternoons in recent memory: a Virtual Safety Car triggered by the prediction engine's own top pick reshuffled the entire strategic picture, the system's second pick never even started, and a driver ranked 18th of 22 stood on the top step.

Here's what the prediction model favored before lights out — and where reality took a dramatically different route. Driver by driver:

George Russell — Predicted P18, Finished P1

No result in this race breaks the prediction algorithm quite like Russell's. The pre-race numbers had him essentially written off at P18, based on a rough recent qualifying average — but they had no way to account for grid position on the actual day, and Russell took pole and converted it into a dominant lights-to-flag victory, Mercedes' first front-row lockout since the 2024 British GP. The real story wasn't just pace: when Hadjar's Red Bull stopped on track and triggered an early Virtual Safety Car, Mercedes reacted instantly and pitted both cars for a "cheap" stop, while Ferrari left its drivers out. That single strategic call decided the race, and Russell never looked back.

Kimi Antonelli — Predicted P4, Finished P2

Antonelli's climb from P4 to P2 tells the same story as his teammate's, just one step behind. He finished 2.9 seconds off Russell after benefiting from the exact same early VSC pit stop that sprang Mercedes clear of the field. It's a strong result relative to his forecast, but more importantly, it confirms the prediction model's projected points average for him (18.0 per race, the highest of anyone in the field) wasn't far off once Mercedes' race-day execution matched the car's pace.

Charles Leclerc — Predicted P5, Finished P3

Leclerc actually out-drove his prediction for most of the afternoon — he jumped Russell into the lead early from P4 on the grid, and the two swapped the lead seven times in the opening ten laps in a genuine scrap. Where the result diverges from the headline story is strategy, not pace: Ferrari's decision to leave Leclerc out around the VSC window, rather than mirror Mercedes' cheap stop, cost him track position he couldn't recover, and he settled for a podium instead of the win his early pace suggested was possible.

Lewis Hamilton — Predicted P19, Finished P4

Like Russell, Hamilton was a driver the prediction system had almost no confidence in, projecting him at P19 off a modest recent qualifying average. He beat that by fifteen places. The explanation mirrors Leclerc's exactly: Hamilton was competitive on outright pace but got caught on the wrong side of Ferrari's stay-out strategy during the VSC, unable to match the Mercedes cars' undercut. A P4 finish from a P19 prediction is still a big overperformance — it's just one shaped by the same strategic call that cost his teammate the win.

Lando Norris — Predicted P21, Finished P5

Norris was projected dead last but one, and instead brought the McLaren home P5 with a clean, incident-free race — a reminder that the prediction algorithm's heavy weighting of recent qualifying and points averages didn't reflect the car's genuine competitiveness at a rules-reset race. That said, finishing 51 seconds off the win shows McLaren wasn't matching Mercedes' pace on the day; this was a solid recovery drive rather than a statement of intent, and it came with a big shadow over it — his teammate never made it to the grid.

Max Verstappen — Predicted P22, Finished P6

Ranked dead last by the prediction engine based on a modest recent points average, Verstappen instead delivered a proper recovery drive. He crashed in Q1 and started all the way down in P20, so P6 represents a genuinely strong afternoon of overtaking rather than a case of the prediction system simply being wrong about pace — it just had no way to know a Saturday crash would define his weekend before it happened.

Oliver Bearman — Predicted P3, Finished P7

Bearman was the prediction model's third pick on the back of a strong recent qualifying average, and he did score points — but four places lower than forecast. He qualified P12, had a difficult opening lap, and only fought back into the top eight by using the second VSC window to switch onto hard tyres, eventually passing Lindblad for P7. Haas called it a needed points "boost," but the underlying story is a tougher qualifying and race than the prediction tool's inputs suggested.

Arvid Lindblad — Predicted P9, Finished P8

About as close to spot-on as this race gets. The 18-year-old debutant — the fourth-youngest driver in F1 history — briefly ran as high as P3 off the start and held on for a points finish that's being talked about as a standout rookie performance, essentially matching his P9 projection almost exactly.

Gabriel Bortoleto — Predicted P6, Finished P9

A slight undershoot relative to a promising forecast built on strong recent qualifying form. Bortoleto still delivered a genuinely good result — Q3 on Saturday and two points on Audi's F1 debut as a manufacturer — just not quite the top-six the prediction system favored, with the sister Audi's dramatic no-show (more on that below) underlining how much better his own weekend went by comparison.

Pierre Gasly — Predicted P20, Finished P10

Gasly comfortably beat his lowly forecast, gaining places at the start from P14 on the grid and grinding out Alpine's first point of the new season. There's no single incident behind this one — it's simply a case of a well-executed, clean race outperforming a prediction framework that had rated his recent form poorly.

Esteban Ocon — Predicted P8, Finished P11

Ocon undershot a promising forecast by three places. He actually ran as high as P8 on lap one and cycled through the first VSC window onto hard tyres just like teammate Bearman, but came out the wrong side of the points cut-off, finishing P11 — a case of the strategic reshuffle around the VSC working out slightly worse for him than for his teammate.

Alexander Albon — Predicted P7, Finished P12

Albon's Saturday was better than his Sunday: he qualified a strong P6, well ahead of the prediction model's expectations, but faded in the race to finish P12, a lap down. The qualifying pace the prediction algorithm had banked on simply didn't translate into race pace on Sunday.

Liam Lawson — Predicted P10, Finished P13

A rough one relative to the forecast. Lawson dropped from P8 on the grid all the way to P18 at the start due to a power and launch problem, and then fought recurring energy-management issues for the rest of the afternoon, recovering only to P13 — reporting suggested continued frustration with the car through team radio, a tougher day than expected.

Franco Colapinto — Predicted P12, Finished P14

Colapinto's race was derailed by a procedural error rather than a pace problem: a 10-second stop-and-go penalty on lap 10, handed down after a mechanic touched the car after the 15-second grid signal, ended any hope of matching or beating his P12 projection and left him two places worse off in P14.

Carlos Sainz — Predicted P14, Finished P15

Almost exactly on the number. Sainz qualified a promising P10 despite nearly missing the session after losing most of FP3, but like teammate Albon he faded in the race, finishing two laps down in P15 — a qualifying strength that again didn't hold up once the race got underway.

Sergio Perez — Predicted P17, Finished P16

Almost a carbon copy of the forecast. Perez brought Cadillac home in P16, a historic first finish for F1's newest team — a solid, low-drama result that lines up with where the prediction model expected him.

Lance Stroll — Predicted P15, Finished P17 (Unclassified)

Stroll finished on track but was classified outside the results, well below his P15 projection. Aston Martin's whole weekend was reportedly compromised by Honda power-unit vibration issues serious enough that both drivers said they risked nerve damage; once out of points contention, the team effectively treated the session as a mileage-gathering exercise, with both cars spending extended spells in the garage rather than racing for position.

Fernando Alonso — Predicted P13, Finished DNF

Alonso's race is one of the strangest results of the day: he was called in to retire on lap 15, "unretired" and rejoined the track, then pitted again on lap 36 before the DNF was made official on lap 37. The underlying cause was the same power-unit vibration problem hampering Stroll — Aston Martin management and technical boss Adrian Newey were reportedly grappling with a car that had missed mileage in pre-season testing, turning a difficult weekend into what was effectively a data-gathering exercise dressed up as a race.

Valtteri Bottas — Predicted P16, Finished DNF

Bottas retired on lap 16 with what's been described as a fuel-system or mechanical issue, well short of his P16 projection. Cadillac's management still framed the overall debut weekend as a positive marker for a brand-new F1 entry, with teammate Perez salvaging the historic first finish.

Isack Hadjar — Predicted P1, Finished DNF

The biggest reversal in the field. Hadjar was the prediction engine's outright favorite off the back of strong recent qualifying and a fantastic Saturday — he qualified P3, with Red Bull boss Laurent Mekies praising a "fantastic" rookie weekend — but retired around lap 10 with a power-unit failure. The knock-on effect was enormous: his stoppage triggered the Virtual Safety Car that Mercedes exploited for their race-winning strategy, meaning the prediction model's top pick didn't just fail to score — he inadvertently decided who did.

Oscar Piastri — Predicted P2, Finished DNS

The prediction system's second-favorite never made the grid. Piastri crashed heavily at Turn 4 on the way to the start, with reporting pointing to a combination of cold tyres, an unforgiving exit kerb, and the car unexpectedly delivering roughly 100kW more power than expected after a pit issue left the battery not fully charged. It's an outcome no pre-race prediction model could have anticipated, and it wiped out what was projected as a near-certain podium.

Nico Hulkenberg — Predicted P11, Finished DNS

Hulkenberg qualified a reasonable P11 but suffered a technical issue on the way to the grid; Audi lost telemetry, couldn't diagnose the fault in time, and withdrew the car before the start. The team reportedly opened an internal investigation afterward — a frustrating end to what should have been a straightforward points chance on the manufacturer's F1 debut weekend, especially with sister car Bortoleto scoring just fine.

Final Takeaways

If there's one lesson from Melbourne, it's that a season-opening race under brand-new regulations is nearly impossible to predict using qualifying and points averages alone — grid position, reliability, and the butterfly effect of a single Virtual Safety Car matter far more than recent form. The prediction algorithm's boldest calls, Hadjar for the win and Piastri for a podium, both ended in a DNF and a DNS respectively, while two of its most written-off drivers, Russell and Hamilton, finished P1 and P4. The Virtual Safety Car triggered by Hadjar's own retirement turned out to be the race's pivot point, rewarding Mercedes' sharp strategy call and punishing Ferrari's decision to stay out. Add in Piastri's freak pre-race crash, Hulkenberg's grid-walk withdrawal, and Aston Martin's power-unit troubles turning their weekend into a test session, and it's clear this opener was defined far more by chaos and reliability than by raw competitive order — a fitting, if unpredictable, start to F1's new era.